import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score, accuracy_score, classification_report, confusion_matrix
import plotly.express as px
from xgboost import XGBClassifier
from imblearn.over_sampling import SMOTE
from sklearn.preprocessing import LabelEncoder

file_path = "air_bnb.csv"
df = pd.read_csv(file_path)

df.head(), df.info(), df.describe(include="all")

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 22552 entries, 0 to 22551
Data columns (total 16 columns):
 #   Column                          Non-Null Count  Dtype  
---  ------                          --------------  -----  
 0   id                              22552 non-null  int64  
 1   name                            22493 non-null  object 
 2   host_id                         22552 non-null  int64  
 3   host_name                       22526 non-null  object 
 4   neighbourhood_group             22552 non-null  object 
 5   neighbourhood                   22552 non-null  object 
 6   latitude                        22552 non-null  float64
 7   longitude                       22552 non-null  float64
 8   room_type                       22552 non-null  object 
 9   price                           22552 non-null  int64  
 10  minimum_nights                  22552 non-null  int64  
 11  number_of_reviews               22552 non-null  int64  
 12  last_review                     18644 non-null  object 
 13  reviews_per_month               18638 non-null  float64
 14  calculated_host_listings_count  22552 non-null  int64  
 15  availability_365                22552 non-null  int64  
dtypes: float64(3), int64(7), object(6)
memory usage: 2.8+ MB

(     id                                              name  host_id host_name  \
 0  2015  Berlin-Mitte Value! Quiet courtyard/very central     2217       Ian   
 1  2695                Prenzlauer Berg close to Mauerpark     2986   Michael   
 2  3176                   Fabulous Flat in great Location     3718    Britta   
 3  3309                 BerlinSpot Schöneberg near KaDeWe     4108      Jana   
 4  7071                  BrightRoom with sunny greenview!    17391    Bright   
 
       neighbourhood_group             neighbourhood   latitude  longitude  \
 0                   Mitte           Brunnenstr. Süd  52.534537  13.402557   
 1                  Pankow  Prenzlauer Berg Nordwest  52.548513  13.404553   
 2                  Pankow   Prenzlauer Berg Südwest  52.534996  13.417579   
 3  Tempelhof - Schöneberg           Schöneberg-Nord  52.498855  13.349065   
 4                  Pankow            Helmholtzplatz  52.543157  13.415091   
 
          room_type  price  minimum_nights  number_of_reviews last_review  \
 0  Entire home/apt     60               4                118  2018-10-28   
 1     Private room     17               2                  6  2018-10-01   
 2  Entire home/apt     90              62                143  2017-03-20   
 3     Private room     26               5                 25  2018-08-16   
 4     Private room     42               2                197  2018-11-04   
 
    reviews_per_month  calculated_host_listings_count  availability_365  
 0               3.76                               4               141  
 1               1.42                               1                 0  
 2               1.25                               1               220  
 3               0.39                               1               297  
 4               1.75                               1                26  ,
 None,
                   id            name       host_id host_name  \
 count   2.255200e+04           22493  2.255200e+04     22526   
 unique           NaN           21873           NaN      5997   
 top              NaN  Berlin Wohnung           NaN      Anna   
 freq             NaN              14           NaN       216   
 mean    1.571560e+07             NaN  5.403355e+07       NaN   
 std     8.552069e+06             NaN  5.816290e+07       NaN   
 min     2.015000e+03             NaN  2.217000e+03       NaN   
 25%     8.065954e+06             NaN  9.240002e+06       NaN   
 50%     1.686638e+07             NaN  3.126711e+07       NaN   
 75%     2.258393e+07             NaN  8.067518e+07       NaN   
 max     2.986735e+07             NaN  2.245081e+08       NaN   
 
              neighbourhood_group         neighbourhood      latitude  \
 count                      22552                 22552  22552.000000   
 unique                        12                   136           NaN   
 top     Friedrichshain-Kreuzberg  Tempelhofer Vorstadt           NaN   
 freq                        5497                  1325           NaN   
 mean                         NaN                   NaN     52.509824   
 std                          NaN                   NaN      0.030825   
 min                          NaN                   NaN     52.345803   
 25%                          NaN                   NaN     52.489065   
 50%                          NaN                   NaN     52.509079   
 75%                          NaN                   NaN     52.532669   
 max                          NaN                   NaN     52.651670   
 
            longitude     room_type         price  minimum_nights  \
 count   22552.000000         22552  22552.000000    22552.000000   
 unique           NaN             3           NaN             NaN   
 top              NaN  Private room           NaN             NaN   
 freq             NaN         11534           NaN             NaN   
 mean       13.406107           NaN     67.143668        7.157059   
 std         0.057964           NaN    220.266210       40.665073   
 min        13.103557           NaN      0.000000        1.000000   
 25%        13.375411           NaN     30.000000        2.000000   
 50%        13.416779           NaN     45.000000        2.000000   
 75%        13.439259           NaN     70.000000        4.000000   
 max        13.757642           NaN   9000.000000     5000.000000   
 
         number_of_reviews last_review  reviews_per_month  \
 count        22552.000000       18644       18638.000000   
 unique                NaN        1312                NaN   
 top                   NaN  2018-11-04                NaN   
 freq                  NaN         618                NaN   
 mean            17.840679         NaN           1.135525   
 std             36.769624         NaN           1.507082   
 min              0.000000         NaN           0.010000   
 25%              1.000000         NaN           0.180000   
 50%              5.000000         NaN           0.540000   
 75%             16.000000         NaN           1.500000   
 max            498.000000         NaN          36.670000   
 
         calculated_host_listings_count  availability_365  
 count                     22552.000000      22552.000000  
 unique                             NaN               NaN  
 top                                NaN               NaN  
 freq                               NaN               NaN  
 mean                          1.918233         79.852829  
 std                           3.667257        119.368162  
 min                           1.000000          0.000000  
 25%                           1.000000          0.000000  
 50%                           1.000000          4.000000  
 75%                           1.000000        129.000000  
 max                          45.000000        365.000000  )

print("Missing Values:\n", df.isnull().sum())
print("Room Type Distribution:\n", df["room_type"].value_counts())
print("Price Description:\n", df["price"].describe())

Missing Values:
 id                                   0
name                                59
host_id                              0
host_name                           26
neighbourhood_group                  0
neighbourhood                        0
latitude                             0
longitude                            0
room_type                            0
price                                0
minimum_nights                       0
number_of_reviews                    0
last_review                       3908
reviews_per_month                 3914
calculated_host_listings_count       0
availability_365                     0
dtype: int64
Room Type Distribution:
 Private room       11534
Entire home/apt    10722
Shared room          296
Name: room_type, dtype: int64
Price Description:
 count    22552.000000
mean        67.143668
std        220.266210
min          0.000000
25%         30.000000
50%         45.000000
75%         70.000000
max       9000.000000
Name: price, dtype: float64

sns.set(style="whitegrid")
fig, axs = plt.subplots(2, 2, figsize=(15, 10))

sns.histplot(df[df['price'] < 500]['price'], bins=50, ax=axs[0, 0], kde=True, color='skyblue')
axs[0, 0].set_title("Distribusi Harga (<500 EUR)")

sns.countplot(x='room_type', data=df, ax=axs[0, 1], palette='Set2')
axs[0, 1].set_title("Jenis Kamar")

sns.histplot(df[df['number_of_reviews'] < 100]['number_of_reviews'], bins=50, ax=axs[1, 0], color='orange', kde=True)
axs[1, 0].set_title("Distribusi Jumlah Review (<100)")

avg_price = df.groupby("neighbourhood_group")["price"].mean().sort_values(ascending=False)
sns.barplot(x=avg_price.values, y=avg_price.index, ax=axs[1, 1], palette="magma")
axs[1, 1].set_title("Rata-rata Harga / Area")

plt.tight_layout()
plt.show()

plt.figure(figsize=(12, 8))
sns.heatmap(df.corr(numeric_only=True), annot=True, cmap='coolwarm')
plt.title("Heatmap Korelasi Numerik")
plt.show()

plt.figure(figsize=(8,6))
sns.scatterplot(data=df[df['price'] < 300], x='number_of_reviews', y='price', alpha=0.5)
plt.title("Harga vs Jumlah Review")
plt.show()

fig = px.scatter_mapbox(df[df['price'] < 500], 
                        lat="latitude", lon="longitude", 
                        color="price", size="availability_365",
                        color_continuous_scale="Viridis",
                        mapbox_style="open-street-map", zoom=10)
fig.update_layout(title="Peta Sebaran Harga Airbnb di Berlin")
fig.show()

df = df.drop(columns=["id", "name", "host_id", "host_name", "last_review"])
df["reviews_per_month"] = df["reviews_per_month"].fillna(0)
df = df[(df["price"] > 0) & (df["price"] < 500)]
df = df[df["minimum_nights"] <= 60]

df["log_price"] = np.log1p(df["price"])
df["long_term"] = (df["minimum_nights"] > 30).astype(int)
df["high_availability"] = (df["availability_365"] > 200).astype(int)
df["review_density"] = df["reviews_per_month"] / df["calculated_host_listings_count"]


df = pd.get_dummies(df, columns=["neighbourhood_group", "neighbourhood", "room_type"], drop_first=True)

X = df.drop(columns=["price", "log_price"])
y = df["log_price"]

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)

rf = RandomForestRegressor(n_estimators=100, random_state=42)
rf.fit(X_train_scaled, y_train)
rf_pred = np.expm1(rf.predict(X_test_scaled))
rf_actual = np.expm1(y_test)

print("\n📊 Evaluasi Random Forest:")
print("MAE:", mean_absolute_error(rf_actual, rf_pred))
print("RMSE:", np.sqrt(mean_squared_error(rf_actual, rf_pred)))
print("R2:", r2_score(rf_actual, rf_pred))

📊 Evaluasi Random Forest:
MAE: 18.823662845337214
RMSE: 33.02665355020505
R2: 0.35815485536925196

gb = GradientBoostingRegressor(n_estimators=200, max_depth=4, learning_rate=0.1, random_state=42)
gb.fit(X_train_scaled, y_train)
gb_pred = np.expm1(gb.predict(X_test_scaled))

print("\n📊 Evaluasi Gradient Boosting:")
print("MAE:", mean_absolute_error(rf_actual, gb_pred))
print("RMSE:", np.sqrt(mean_squared_error(rf_actual, gb_pred)))
print("R2:", r2_score(rf_actual, gb_pred))

📊 Evaluasi Gradient Boosting:
MAE: 18.71246739143494
RMSE: 33.48381325569928
R2: 0.340262848384806

df["price_class"] = pd.cut(df["price"], bins=[0, 50, 100, 500], labels=["Murah", "Sedang", "Mahal"])

X_class = df.drop(columns=["price", "log_price", "price_class"])
y_class = df["price_class"]

X_train_c, X_test_c, y_train_c, y_test_c = train_test_split(X_class, y_class, test_size=0.2, random_state=42)
X_train_c_scaled = scaler.fit_transform(X_train_c)
X_test_c_scaled = scaler.transform(X_test_c)

le = LabelEncoder()
y_train_enc = le.fit_transform(y_train_c)
y_test_enc = le.transform(y_test_c)

smote = SMOTE(random_state=42)
X_train_sm, y_train_sm = smote.fit_resample(X_train_c_scaled, y_train_enc)

clf = XGBClassifier(use_label_encoder=False, eval_metric="mlogloss", random_state=42)
clf.fit(X_train_sm, y_train_sm)

y_pred_enc = clf.predict(X_test_c_scaled)

y_pred_label = le.inverse_transform(y_pred_enc)
y_test_label = le.inverse_transform(y_test_enc)

print("\n📊 Evaluasi XGBoost Klasifikasi:")
print("Accuracy:", accuracy_score(y_test_label, y_pred_label))
print("Classification Report:\n", classification_report(y_test_label, y_pred_label))

/opt/anaconda3/lib/python3.11/site-packages/xgboost/training.py:183: UserWarning:

[08:56:51] WARNING: /Users/runner/work/xgboost/xgboost/src/learner.cc:738: 
Parameters: { "use_label_encoder" } are not used.

📊 Evaluasi XGBoost Klasifikasi:
Accuracy: 0.6975687343785504
Classification Report:
               precision    recall  f1-score   support

       Mahal       0.38      0.39      0.38       379
       Murah       0.82      0.81      0.82      2618
      Sedang       0.56      0.57      0.56      1404

    accuracy                           0.70      4401
   macro avg       0.59      0.59      0.59      4401
weighted avg       0.70      0.70      0.70      4401

A. Analisis 1. Distribusi Harga (Gambar 1 kiri atas) Mayoritas listing memiliki harga <100 EUR/malam. Harga sangat right-skewed, menandakan ada banyak listing murah dan sedikit yang mahal. 2. Jenis Kamar (kanan atas) Dominasi Private Room dan Entire home/apt. 3. Jumlah Review (kiri bawah) Mayoritas listing punya review sedikit menunjukkan bahwa banyak listing baru/kurang populer. Korelasi terhadap harga cenderung lemah. 4. Rata-rata Harga per Area (kanan bawah) Area mahal: Charlottenburg-Wilmersdorf, Tempelhof-Schöneberg, Mitte Area murah: Neukölln, Reinickendorf, Treptow-Köpenick 5. Harga vs Jumlah Review (scatterplot) Tidak ada hubungan kuat antara harga tinggi dengan banyak review ada hipotesa lebih penting (lokasi, ketersediaan, dll). Harga rendah cenderung membuat banyak user melakukan review 6. Peta Sebaran Harga Airbnb Harga tinggi tersebar di tengah kota. Terdapat konsentrasi listing dengan harga murah di pinggiran. B. Pengolahan data 1. Model Regresi – Prediksi Harga Model MAE RMSE R² Analisis Random Forest 18.82 33.02 0.36 Baik, tapi masih underfit. Gradient Boosting 18.71 33.48 0.34 Mirip RF, perbedaan kecil. Insight: R² < 0.4 → model belum menangkap cukup banyak variasi harga. Harga yang sangat skewed ditambah fitur terbatas membuat model kesulitan belajar. C. Rekomendasi & Next Steps 1. Model & Data: Gunakan model XGBoost atau CatBoost untuk regresi. Tambahkan data eksternal: kalender musim/liburan, lokasi wisata. Gunakan teknik feature selection untuk kurangi noise. 2. Business Insight: Platform bisa memberi rekomendasi harga otomatis bagi host baru berdasarkan area dan fitur. Identifikasi listing dengan potensi kenaikan harga berdasarkan lokasi dan demand historis. Lokasi dengan harga murah tapi demand tinggi bisa jadi target promosi. Kesimpulan Singkat: Telah melakukan eksplorasi data Airbnb dan membangun dua model machine learning. Model regresi memiliki akurasi MAE 18 EUR dan R2 0.35. Model klasifikasi berhasil mengenali tiga kategori harga (Murah, Sedang, dan Mahal) dengan akurasi sekitar 70%. Setelah diterapkan teknik balancing menggunakan SMOTE, performa model meningkat secara signifikan khususnya dalam mengenali kategori “Mahal” yang sebelumnya gagal dikenali. Hasil visualisasi memperkuat insight bahwa harga sangat dipengaruhi oleh lokasi geografis dan tipe kamar. Namun, jumlah review tidak berbanding lurus dengan harga, justru dengan harga lebih murah cenderung mendapatkan lebih banyak ulasan hipotesa karena lebih banyak disewa oleh wisatawan dengan anggaran terbatas (untuk hal ini perlu diihat distribusi reviewnya negatif atau positif) jika saya hitung nilai korelasi Pearson antara price dan number_of_reviews, hasilnya kemungkinan negatif dan kecil (mungkin sekitar -0.3). oleh karena itu tidak bisa dibilang hubungan sebab-akibat (bisa jadi listing murah lebih sering disewa karena affordable, lalu otomatis dapat lebih banyak review atau karena harga murah mendapakan fasiliast yang kurang baik).